Paternal inheritance of the primary sex ratio in a copepod

(1)

Paternal inheritance of the primary sex ratio in a copepod

M. J. VOORDOUW,* H. E. ROBINSON & B. R. ANHOLT*

*Department of Biology, University of Victoria, Victoria, British Columbia, Canada Department of Marine Science, University of Texas, Austin, Texas, USA

Introduction

In sexually reproducing organisms, biparentalism is the cornerstone of Fisher’s sex-ratio principle (Bull &

Charnov, 1988). From the perspective of an autosomal gene, males and females are equally efficient means of getting to the next generation and the optimal sex ratio is therefore 1 : 1 (Fisher, 1930). In contrast, genetic elements that are predominantly transmitted through either males or females are under strong selection to skew the primary sex ratio towards the transmitting sex (Lewis, 1941; Howard, 1942; Hamilton, 1967). Theory suggests that these asymmetrically inherited sex-ratio distorters can drive their host population to extinction (Hamilton, 1967; Cosmides & Tooby, 1981; Taylor, 1990) and populations with extremely biased sex ratios have been observed (Juchaultet al., 1993; Jigginset al., 2000). Their

‘selfish nature’ is a striking illustration that not all genes operate in the organism’s best interests (Dawkins, 1976).

Sex-ratio distorters include sex chromosomes with meiotic drive (Jaenike, 2001), maternally inherited

microorganisms (O’Neil et al., 1997; Werren, 1997;

Stouthameret al., 1999; Weeks et al., 2002) and supernumerary ‘B’ chromosomes (Werren & Stouthamer, 2003). Some populations of fruitflies have an X chromosome with meiotic drive that causes males (the heterogametic sex) to produce predominantly X-bearing sperm resulting in a female-biased sex-ratio (Varandas et al., 1997; Carvalhoet al., 1998). In several species of isopod, a cytoplasmically (maternally) inherited bacterium, Wolbachia, feminizes genetic males into phenotypic females (Rigaud, 1997). In two species of parasitoid wasp, Nasonia vitripennis and Trichogramma kaykai, a small, paternally transmitted B chromosome converts diploid female embryos into haploid male embryos by destroying the other paternal chromosomes (Werren &

Stouthamer, 2003). In each of the above examples, an asymmetric inheritance pattern selects for genetic elements that skew the primary sex ratio towards the transmitting sex.

Traditional lineage analysis of family sex ratios is often the first step in establishing the presence of a uniparentally inherited sex-ratio distorter (Hurst, 1993). For example, the maternal inheritance of a feminizing genetic element in the isopod, Armadillidium vulgare, had been characterized for several decades (Legrand et al., 1987) prior to its molecular identification as the

Correspondence:M. J. Voordouw, Department of Biology, University of Victoria, P.O. Box 3020, Victoria, British Columbia, Canada V8W 3N5.

Tel.: +1-250 472-4684; fax: +1-250 721-7120;

e-mail: voordouw@uvic.ca

Keywords: copepod; heritability; maternal transmission; paternal transmission; primary sex ratio;Tigriopus californicus.

Abstract

Uniparentally inherited genetic elements are under strong selection to manipulate sex determination in their host and shift the host sex ratio towards the transmitting sex. For any sex-ratio trait, lineage analysis and quantitative genetics are important tools for characterizing the mode of inheritance (biparental vs. maternal vs. paternal) thereby narrowing the field of possible sex-determining mechanisms (e.g. polygenic, sex chromosomes with meiotic drive, cytoplasmic microorganisms). The primary sex ratio of the harpacticoid copepod,Tigriopus californicusis often male-biased and is highly variable among full sib families. We found that this extra-binomial variation for the primary sex ratio is paternally but not maternally transmitted in T. californicus. Paternal transmission of the primary sex ratio has been well documented in the haplo–diploid hymenoptera but is relatively rare in diplo–

diploid organisms. If the sex-ratio trait is paternally transmitted in other closely related harpacticoid copepods it would explain why male biased primary sex ratios are so common in this group.

(2)

bacterium Wolbachia(Rousset et al., 1992). Likwise, the paternal sex-ratio (PSR) factor of N. vitripennis was accidentally discovered during a selection experiment (Werren et al., 1981) and its mechanism was subsequently elucidated (Werren & Stouthamer, 2003). While lineage experiments cannot pinpoint the mechanism per se, the inheritance pattern (i.e. maternal vs. paternal) can narrow the field of potential candidates. For example, a paternally transmitted sex-ratio trait would eliminate Wolbachia (or other maternally inherited microorganisms) and would suggest the presence of a male heterogametic system with Y-drive.

Over the last 5 years, we have studied the sex-ratio trait in the harpacticoid copepod, Tigriopus californicus.

We have repeatedly found that this splash-pool copepod has a large among family component in the primary sex ratio; i.e. full sibs covary in their tendency to develop into a male (Voordouw & Anholt, 2002b).

Experiments investigating environmental sex determination (ESD) suggest that temperature (Voordouw &

Anholt, 2002a) and density (unpublished data) do not play a significant role in structuring the sex-ratio variance inT. californicus. The sex-determining mechanism for T. californicus remains unknown. Karyotypes have found no heteromorphic sex chromosomes (Ar-Rushdi, 1963) and several authors have claimed a polygenic sex-determining mechanism (Ar-Rushdi, 1958; Belser, 1959; Egloff, 1966; Voordouw & Anholt, 2002b).

We reported a parent-offspring correlation in the sex ratio between mothers and daughters but this experiment had poor survivorship in both generations so that we cannot rule out the presence of sex-specific mortality (Voordouw & Anholt, 2002b). In addition, the mother- offspring regression is potentially confounded by maternal effects which is not a problem for the father-offspring regression (Falconer, 1989). The recent demonstration

that the primary sex ratio is paternally inherited in the European fairy shrimp, Branchipus schaefferi (Beladjal et al., 2002), motivated us to investigate the paternal contribution to the sex-ratio trait in T. californicus. Like B. schaefferi, we found that the sex-ratio trait was paternally transmitted in T. californicus, although the mechanism that generates this inheritance pattern is currently unknown.

Materials and methods

General overview

The lineage experiment consists of three generations: the F1, the F2and F3 generation (Fig. 1). There are seven different types of relatives and all of them are defined in relation to the F3generation. The F1generation consists of the paternal grandfathers, the paternal grandmothers, the maternal grandfathers and the maternal grandmothers. The F2generation consists of the fathers and mothers and the F₃generation consists of offspring.

Primary sex ratio of a family

The protocol for determining the primary sex ratio of a family is as follows. We removed the egg sac from a gravid female and allowed it to hatch in a glass spot plate. Shortly after hatching, we haphazardly sampled 24 nauplii from the spot plate. We reared these nauplii to sexual maturity (about 2 weeks under laboratory conditions); at which time it is easy to differentiate between males and females.

Rearing protocol

Each nauplius was reared in its own well in a 24-well tissue culture plate. Each well was stocked with 2.5 mL

Paternal grandfather (n = 112)

Paternal grandmother (n = 80)

Maternal grandfather (n = 106)

Maternal grandmother (n = 106)

Father (n = 72)

Mother (n = 88)

Offspring (n = 70)

F₁

F₂

F₃

May 01 May 20 May 22 June 03

June 10 June 24

July 15 Fig. 1Experimental design of the lineage

experiment showing the three generations (F1, F2 and F3), the seven relatives (paternal grandfathers, paternal grandmothers, maternal grandfathers, maternal grandmothers, fathers, mothers and offspring), the approximate dates on which each relative was born and the sample sizes (n).

(3)

of filtered sea water, 4.0 · 10⁶cells of Isochrysis galbaenaand0.2 mg of Tetramin flakes. Tissue culture plates were stored on top of moist paper towels in covered plastic containers to prevent evaporation.

Containers were stored in incubators set at a constant temperature of 20C with a 12 h day/night cycle.

Field collection

Like many other crustaceans, T. californicus exhibits precopulatory mate-guarding. Under natural mating conditions, adult males typically clasp sexually immature females (copepodite stages I and II) and guard them until they reach sexual maturity (copepodite stage VI) at which time mating occurs (Burton, 1985). To accommo- date this precopulatory mate-guarding behaviour, the field collection and rearing of the four grandparents were staggered in time.

We collected100 gravid females from the field on each of four sampling dates (1 May, 20 May, 22 May and 3 June 2003) and used the egg sacs from these field- captured females to create the paternal grandfathers (n¼112), the paternal grandmothers (n¼80), the maternal grandfathers (n¼106) and the maternal grandmothers (n¼106), respectively (Fig. 1). The 19-day delay (1 May–20 May) between the paternal grandfathers and grandmothers and the 12-day delay (22 May–3 June) between the maternal grandfathers and grandmothers ensured that the sires reached adulthood well before the dams. Likewise, the 14-day delay (20 May–3 June) between the paternal and maternal grandmothers ensured that the F2fathers reached adulthood prior to the F2mothers.

Mating protocol

The paternal F1cross refers to the mating of the paternal grandparents, the maternal F₁cross refers to the mating of the maternal grandparents, and the F2cross refers to the mating of the fathers and mothers (Fig. 1). For all three crosses (paternal F1, maternal F1and F2), we used a random number generator to pair the families from which the sires and dams were subsequently selected (random pairing).

F1mating protocol

For each randomly assigned pair of sire and dam families, we used a mating arena to introduce three haphazardly selected adult males (brothers A, B and C) from the sire family to 12 full-sib copepodites of unknown sex phenotype from the dam family. The mating arena consisted of a single well in a 6-well tissue culture plate stocked with 5 mL of filtered sea water and 5–10 million cells ofI. galbaena. Once a sexually mature male from the sire family (i.e. a future grandfather) had clasped a

female copepodite from the dam family (i.e. a future grandmother), we removed that couple from the mating arena. We moved the remaining, unclasped copepodites from the mating arena to a 24-well tissue culture plate and reared them to sexual maturity to estimate the dam family sex ratio. The F1 mating protocol simulated natural courtship in that all female dams were guarded by the sires prior to mating. However, this protocol was very time consuming and so we changed it for the F2

cross.

F2mating protocol

In the F₂ cross, we reared all sire and dam families to sexual maturity prior to mating them. For each randomly assigned pair of sire and dam families, we haphazardly selected two or three adult males from the sire family and introduced them to two or three haphazardly selected adult females from the dam family. In this mating protocol there was no need for a mating arena.

Biological differences between the F1and the F2

cross

In the paternal and maternal F₁ crosses, the males engaged in precopulatory mate-guarding for several days prior to mating. In contrast, for the F₂crosses, there was no precopulatory mate-guarding and mating occurred immediately following introduction. Furthermore, in the F1mating arenas, the males had to search for and select a female copepodite while competing with their two brothers. No such sexual selection occurred in the F2

crosses.

Sample size considerations and nonindependence of crosses

Over the course of the experiment, the sample size inevitably shrunk because some crosses failed to produce offspring. To combat this problem, we often kept two or three replicate couples (A, B and C) for a given cross (e.g. sire family 35* dam family 201). For the replicate couples 35A*201A and 35B*201B, the sires (35A and 35B) are full sib brothers (from family 35), the dams (201A and 201B) are full sib sisters (from family 201) and their offspring will be double first cousins. The genetic coefficient of relatedness among double first cousins is 1/4.

In some cases, we used related individuals to increase the sample size in the next generation. The disadvantage of this approach is that it increases the genetic relatedness among individuals over successive generations. Of the 79 F2 crosses, four involved fathers who were full sib brothers (r¼1/2) and five involved fathers who were double first cousins (r¼1/4). The genetic coefficient of relatedness among F3 families whose fathers were full

(4)

sibs or double first cousins is 1/8 and 1/16, respectively.

In the results we account for this genetic covariance by aggregating the sex phenotype data across paternal grandfathers. Hence the average number of offspring per family is often greater than 24 individuals.

For the paternal F1 cross, 72 of the 80 couples (112 paternal grandfathers*80 paternal grandmothers) produced F2 offspring (the fathers). For the maternal cross, 88 of the 106 couples (106 maternal grandfathers*106 maternal grandmothers) produced F2 offspring (the mothers). For the F2 cross (after aggregating the four and five F2 crosses whose fathers were full sibs and double first cousins, respectively), 70 of the 72 couples (72 fathers*88 mothers) produced F3 offspring (Fig. 1).

Statistical methods Larval mortality correction

As we have done previously (Voordouw & Anholt, 2002a,b), we used the larval mortality correction to account for sex-biased mortality. For each plate of full sibs, unidentified individuals were assigned to the less common sex for that plate. The larval-mortality correction tends to reduce the variation in the primary sex ratio and is therefore statistically conservative. Throughout this paper we will refer to both the uncorrected (raw) and the larval-mortality corrected data.

Mean and variance in the proportion of males

For each relative, we used binomial randomization tests to determine whether the observed mean and variance in the proportion of males was significantly different from the Mendelian expectation (i.e. independent assortment of sex chromosomes). For each parent – offspring pair, we used a bootstrap procedure to determine the statistical significance of any differences in the sex-ratio mean and variance between parents and offspring.

To compare the distribution of the proportion of males among relatives, we only included those 70 families that actually produced F3offspring. Excluded families include those families that were not selected in the mating protocol and those families that failed to produce either sons or daughters due to low survivorship and/or extreme sex ratios. Excluding the latter type of family (low survivorship and/or extreme sex ratios) will tend to reduce the variance and/or the skew in the population sex ratio (i.e. if sex-determination and/or larval-mortality is sex-biased). Our estimates of the sex-ratio mean and sex-ratio variance for each population of relatives should therefore be conservative.

Paternal inheritance of the primary sex ratio

To determine whether fathers have a stronger effect than mothers on the primary sex ratio of their offspring (orvice versa), we used multiple regression to compare the partial regression coefficients of the father – offspring and the

mother – offspring regressions for all three crosses (F₁ paternal cross, F1maternal cross, F2cross) separately and combined.

To compare this study to Voordouw & Anholt (2002b)), we treated sex as a quantitative threshold trait (Bulmer & Bull, 1982) and estimated the heritability of sex tendency following the method by Roff (1986,1997) with one important exception. For an individual from the ith family, Roff (1986,1997) calculates the phenotypic value of that individual (on the underlying scale) as the ordinate on the standardized normal curve that corresponds to p_i, the proportion of males in that family. This calculation is wrong because families with 10% or 90% males both have the same probability density (0.175). We used the cumulative standardized normal curve instead so that the above two families have a cumulative probability density of 0.035 and 0.965, respectively.

To estimate the heritability of sex tendency (h²), we doubled the slope (b) and standard error from each of the six parent – offspring regressions: (1) paternal grandfather – F2 offspring, (2) paternal grandmother – F2 offspring, (3) maternal grandfather – F2 offspring, (4) maternal grandmother F2 offspring, (5) F2 father – F3 offspring, (6) F2 mother – F3 offspring (for the three midparent – offspring regressions, b¼h²). Again, we only included those 70 families that actually produced F3

offspring.

Results

Survivorship, male-biased mortality and the larval mortality correction

For the seven types of relatives across three generations we reared 18 094 individuals from 608 different families, sexed 15 754 adults (average survivorship¼87.1%) of which 8179 individuals were males (proportion of males¼0.519). For the seven types of relatives, survivorship ranged between 78.5 and 97.9% (Table 1). Sur- vivorship was lowest for the F3 offspring (78.5%) and the maternal grandmothers (81.0%). For the maternal grandmothers, the correlation between survivorship and the raw proportion of males was positive (r¼ 0.485) and highly statistically significant (P< 0.001) suggesting that mortality was male-biased in this relative.

The correlation between the raw and the larval-mortality corrected proportion of males was lowest for the maternal grandmothers (r¼0.744) and this is expected under low survivorship and/or sex-biased mortality. For all other relatives, there was little or no evidence that mortality was sex-biased and the correlation between the raw and corrected proportion of males was high (Table 1). Hence, we are confident that our results are not biased by sex-linked lethal alleles and that they reflect the segregation of genetic factors that affect the primary sex ratio.

(5)

Mean proportion of males within generations For the seven relatives, the mean proportion of males for the raw data was generally similar to that of the larval-mortality corrected data (Table 2). The only exception was the maternal grandmothers where – due to a combination of low survivorship and male-biased mortality (Table 1) – the larval mortality correction increased the proportion of males from 0.33 to 0.42. In either case, the mean proportion of males for the maternal grandmothers was significantly female-biased

whereas it was significantly male-biased for all other grandparents. The sex ratio was not significantly different from 0.5 for the fathers, female-biased for the mothers and male-biased in the F3 offspring (Table 2).

The observation that the primary sex ratio was significantly male-biased for four of the seven relatives is consistent with the idea of a paternally transmitted sex- ratio factor that has drive relative to the rest of the genome.

Mean proportion of males among generations Multiple independent two-samplet-tests found that the balanced sex ratio in the F₂fathers was not significantly different from the male-biased sex ratio in either the paternal grandfathers or the paternal grandmothers (Table 3). In contrast, the mildly female-biased sex ratio in the F2mothers was significantly different from both the highly male-biased sex ratio of the maternal grandfathers and the highly female-biased sex ratio of the maternal grandmothers (Table 3). Finally the proportion of males in the F3offspring was significantly more male- biased than either that of their F₂ mothers or their F₂ fathers (Table 3). Hence, the bias in the primary sex ratio decreased from the F1to the F2generation and increased slightly from the F₂to the F₃generation (Fig. 2). From our random mating protocol, we expect the primary sex ratio to fluctuate at random around the mean of 0.500.

With only two observations it is impossible for us to determine whether the fluctuations are random or directional.

Table 1.Mean survivorship (expressed as percent) ± standard error, the correlation (r) between the raw proportion of males and survivorship and the correlation between the raw and larval- mortality corrected proportion of males for all seven types of relatives. Also shown are the sample size (n) and the statistical significance (P) of the two correlations. Statistically significant correlations are outlined in boldface type.

Relative n

Survivorship

± SE

Raw sex ratio vs.

survivorship

Corrected vs. raw sex ratio

r P r P

Paternal grandfather 112 85.3 ± 1.43 0.175 0.066 0.884 <0.001 Paternal grandmother 80 96.4 ± 0.88 0.023 0.843 0.978 <0.001 Maternal grandfather 106 97.9 ± 0.36 0.200 0.040 0.991 <0.001 Maternal grandmother 106 81.0 ± 2.16 0.485 <0.001 0.744 <0.001 F2father 72 89.8 ± 2.19 –0.065 0.590 0.948 <0.001 F2mother 88 90.8 ± 0.99 0.157 0.144 0.963 <0.001 F3offspring 70 78.5 ± 1.92 –0.081 0.508 0.899 <0.001

Table 2.The mean proportion of males (P.male) ± SE and the observed variance in the proportion of males (Obs. Var.) for all seven types of relatives. We only included those families that actually produced F3 offspring (n¼70 for each relative). We conducted randomization tests to determine if (1) a relative’s mean proportion of males was significantly different from 0.500 (H₀:P.male¼0.50) and (2) whether the observed variance (Obs. Var) in the proportion of males was significantly greater than the expected variance (Exp. Var) under Mendelian segregation of sex chromosomes (H₀: Obs. Var.£Exp. Var.). The ratio is the observed variance in the proportion of males divided by the expected variance. For each type of relative, the top and bottom row show the raw and larval mortality-corrected (LMC) data, respectively (as indicated in the ‘Data’ column). Means and variances that are signifcantly different from the Mendelian expectation are shown in boldface type.

Relative n Data

H0:P.male¼0.50 H0: Obs. Var.£Exp. Var.

P.male ± SE P Obs. Var. Exp. Var. Ratio P

Paternal grandfather 70 Raw 0.55±0.020 <0.001 0.028 0.011 2.5 <0.001

LMC 0.55±0.017 <0.001 0.020 0.010 1.9 <0.001

Paternal grandmother 70 Raw 0.56±0.019 <0.001 0.025 0.011 2.3 <0.001

LMC 0.55±0.018 <0.001 0.022 0.010 2.1 <0.001

Maternal grandfather 70 Raw 0.60±0.025 <0.001 0.042 0.010 4.2 <0.001

LMC 0.59±0.024 <0.001 0.039 0.010 3.9 <0.001

Maternal grandmother 70 Raw 0.33±0.021 <0.001 0.032 0.013 2.5 <0.001

LMC 0.42±0.015 <0.001 0.015 0.009 1.6 <0.001

F2 father 70 Raw 0.51 ± 0.024 0.497 0.039 0.012 3.2 <0.001

LMC 0.50 ± 0.021 0.799 0.030 0.010 3.0 <0.001

F2 mother 70 Raw 0.46±0.024 <0.001 0.041 0.010 4.2 <0.001

LMC 0.47±0.021 0.016 0.030 0.009 3.4 <0.001

F3 offspring 70 Raw 0.58±0.021 <0.001 0.030 0.008 3.8 <0.001

LMC 0.54±0.014 <0.001 0.015 0.006 2.5 <0.001

(6)

Variance in the proportion of males within generations

For the raw data, the observed variance in the proportion of males was, on average, 3.2 times larger than the Mendelian expectation (Table 2; range: 2.3–4.2). For the larval-mortality corrected proportion of males, the average ratio of the observed to the expected variance was 2.6 (Table 2: range:1.6–3.9). Regardless of whether the data were corrected for larval mortality, the observed variance in the primary sex ratio was always significantly greater than the binomial expectation (Table 2). The observation that the variance in the primary sex ratio is conserved across generations is consistent with our random mating protocol (i.e. random mating does not exert any directional selection on the primary sex ratio).

Variance in the proportion of males among generations

For the uncorrected data, multipleF-tests found that the change in sex-ratio variance between parents and offspring was not statistically significant for any of the six parent–offspring pairs (Table 3). For the larval-mortality corrected data, the sex-ratio variance in F2mothers is almost twice as great as that of the maternal grandmothers. Conversely, the variance in F3offspring is half as great as that of their F₂fathers and mothers (Table 3).

In both cases, this change in variance between generations appears to be the result of differences in survivorship (Table 1; the maternal grandmothers and the F3

offspring both had low survivorship, compared to their F2

offspring and F₂parents, respectively). Hence, there do

not appear to be any meaningful changes in sex-ratio variance between generations suggesting, again, that it was conserved in this experiment.

Paternal inheritance of the primary sex ratio

For the paternal F1cross (larval mortality corrected data), the partial regression coefficient for the paternal grandfathers (b¼0.41 ± 0.143) is highly statistically significant (t¼2.842,P¼0.006) and is much larger than the partial regression coefficient for the paternal grandmothers (b¼–0.01 ± 0.137) which is not statistically significant (t¼–0.063,P¼0.950). For the maternal F1cross (larval mortality corrected data), the highly significant partial regression coefficient for the maternal grandfathers (b¼0.34 ± 0.100, t¼3.379,P¼0.001) is almost eight times larger than the nonsignificant partial regression coefficient for the maternal grandmothers (b¼ 0.04 ± 0.159,t¼0.272,P¼0.787). For the maternal F₂ cross (larval mortality corrected data), the partial regression coefficient for the fathers (b¼0.08 ± 0.088) is not statistically significant (t¼0.922, P¼0.360) but still larger than the nonsignificant partial regression coefficient for the mothers (b¼–0.02 ± 0.088, t¼–0.207, P¼0.837). When all three crosses are combined (larval mortality corrected data), the highly significant partial regression coefficient for the father (b¼0.22 ± 0.062, t¼3.463,P¼0.001) was almost eight times larger than that of the mother (b¼ 0.03 ± 0.068, t¼0.410, P¼ 0.682). Hence fathers appear to have a stronger effect than mothers on the offspring sex ratio.

The estimates of the heritability of sex tendency were calculated from parent–offspring regressions (i.e. not Table 3.Comparing the sex-ratio mean and the sex-ratio variance between parents and offspring. We only included those families that actually produced F3 offspring (n¼70 for each parent–offspring pair). For each of the six parent–offspring pairs, we compared the sex-ratio mean between parents and offspring using an independent two-samplet-test (t₁₃₈,P₁) and a bootstrap test (P₂). We compared the sex-ratio variance between parents and offspring using anF-test (F_69,69,P₃) and a bootstrap test (P₄). For each type of relative, the top and bottom row show the raw and larval mortality-corrected (LMC) data, respectively (as indicated in the ‘Data’ column). Statistically significantt-tests and F-tests are shown in boldface type.

Parent Offspring Data

H0:lparent¼loffspring H0:r²_parent¼r²_offspring t-test

P2

F-test

P4

t138 P1 F69,69 P3

Paternal grandfathers F2fathers Raw 1.364 0.175 0.175 1.406 0.159 0.103

LMC 1.671 0.097 0.097 1.513 0.088 0.098

Paternal grandmothers F2fathers Raw 1.646 0.102 0.101 1.577 0.061 0.039

LMC 1.826 0.070 0.070 1.380 0.184 0.198

Maternal grandfathers F2mothers Raw 3.970 <0.001 <0.001 0.957 0.857 0.809

LMC 3.888 <0.001 <0.001 0.769 0.278 0.194

Maternal grandmothers F2mothers Raw –4.158 <0.001 <0.001 1.282 0.305 0.265

LMC –2.252 0.026 0.029 1.961 0.006 0.016

F2fathers F3offspring Raw –2.330 0.021 0.024 0.754 0.244 0.171

LMC –1.621 0.107 0.109 0.490 0.004 0.010

F2mothers F3offspring Raw –3.850 <0.001 <0.001 0.728 0.191 0.109

LMC –2.828 0.006 0.005 0.487 0.003 0.008

(7)

from the multiple regression involving both parents) and reflect the pattern of transmission of the sex-ratio trait (Fig. 3; Table 4). The heritability estimates are high for the paternal grandfather – F₂offspring regression and the maternal grandfather – F2offspring regression. For the F2

father – F3 offspring regression, the heritability of sex tendency is not significantly different from zero, however, if we exclude just two families (F2fathers 137 and 151), the heritability from the corrected data is significantly different from zero (h²¼0.36 ± 0.173), although the heritability from the raw data is not (h²¼ 0.07 ± 0.218). The F2 fathers 137 and 151 (marked in black in Fig. 3) were taken from highly female-biased families (two males per 24 adults and three males per 23 adults) but produced highly male-biased families (18 males per 20 adults and 19 males per 22 adults).

While we have noa priorijustification for removing these two families, it is worthwhile to consider their impact on the F2father – F3offspring heritability estimate. For the mother – offspring regressions, none of the heritability estimates were significantly different from zero (Table 4).

The midparent – offspring heritability was statistically significant for both F₁crosses (due to the large contribution of the grandfathers) but not for the F2cross. Hence

the sex-ratio trait inT. californicuswas clearly paternally inherited in the F₂ generation but not in the F₃ generation.

Discussion

The strength of our results lies in the large sample size (>18 000 individuals), high survivorship (>87%) and our conservative use of the larval mortality correction, which accounts for sex-biased mortality (especially in the maternal grandmothers in Tables 1 and 2). The three major findings of this study are: (1) the mean primary sex ratio is often biased (usually towards males;

Table 2) and can change significantly between generations (i.e. maternal grandparents vs. F2 mothers in Table 3), (2) the observed variance in the primary sex ratio is always larger than the binomial expectation (Table 2) and is similar among generations (Table 3) and (3) this extra-binomial variation in the primary sex ratio appears to be paternally transmitted (at least from F1to F2).

In the present study, the primary sex ratio in two successive field samples dropped from 0.60 ± 0.025 (maternal grandfathers) to 0.33 ± 0.021 (maternal 30

20 10 0

30 20 10 0

30 20 10 0 30

20 10 0

30 20 10 0

1A 1B 1C 1D

2A 2B

1.0 0.8 0.6 0.4 0.2 –0.0

1.0 0.8 0.6 0.4 0.2 –0.0 1.0

0.8 0.6 0.4 0.2 –0.0

1.0 0.8 0.6 0.4 0.2 –0.0 1.0 0.8 0.6 0.4 0.2 –0.0 1.0 0.8 0.6 0.4 0.2 –0.0 1.0 0.8 0.6 0.4 0.2 –0.0

3A

Proportion of males Count

Fig. 2 Distribution of the raw proportion of males for three generations (F1, F2and F3) and seven relatives (1A: paternal grandfathers; 1B:

paternal grandmothers; 1C: maternal grandfathers; 1D: maternal grandmothers; 2A: F2fathers; 2B: F2mothers; 3A: F3 offspring). We only included those families that actually produced F3 offspring (n¼70 for each relative). For each relative, the mean proportion of males (uncorrected for larval mortality) is depicted by the bold-black line.

(8)

grandmothers) in less than 2 weeks and changed to 0.46 ± 0.024 in the next generation (F₂mothers). In the summer of 2002, the mean primary sex ratio in a sample of 47 field-captured females and 167 of their lab-reared daughters changed from 0.59 ± 0.027 to 0.46 ± 0.015 (unpublished data). We have previously suggested that the sex phenotype in adult copepods is influenced by environmental factors operating during naupliar devel- opment (Voordouw & Anholt, 2002a). However, the fact that all of these copepods were reared under similar conditions (i.e. 20C, 12 h light : 12 h dark cycle, abundant I. galbaena food source) argues against this interpretation.

The primary sex ratio is often male-biased in T. californicus. In Voordouw & Anholt (2002a), all eight independent estimates of the primary sex ratio (uncorrected for larval mortality) were ‡0.50 (range: 0.50–

0.68). In a recent experiment where we manipulated larval density, the mean primary sex ratio in the low (0.71 ± 0.027) and high (0.64 ± 0.030) density treat- ment was highly male-biased (unpublished data). In the present study, male bias was less obvious but still more common than female bias (Table 2). Igarashi (1963b) reported highly male-biased sex ratios (range: 0.50–0.91) among families in the closely related T. japonicus. How- ever, his practice of excluding any family with less than

0.0 0.5 1.0 0.0 0.5 1.0 0.0 0.5 1.0

Parental proportion of males 0.0

0.5 1.0

0.0 0.5 1.0

Offspring proportion of males

Paternal grandfather M aternal grandfather Father

Paternal grandmother M aternal grandmother M other

Fig. 3The relationship in the raw proportion of males for the six parent–offspring pairs (paternal grandfathers: F2 fathers;

maternal grandfathers: F2 mothers; F2 fathers: F3 offspring; paternal grandmothers:

F2 fathers; maternal grandmothers: F2 mothers; F2 mothers: F3 offspring). We only included those families that actually produced F3 offspring (n¼70 for each parent–offspring pair). Shown is the line of best fit from the linear regression of the offspring proportion of males vs. the parental proportion of males.

Table 4.The relationship in the proportion of males between parents and offspring. We only included those families that actually produced F3 offspring (n¼70 for each parent–offspring pair). For each of the six parent–offspring regressions, we show the correlation coefficient (r), theF-statistic (F1,68), the statistical significance (P) and the heritability of sex tendency and its standard error (h²± SE). For each type of relative, the top and bottom row show the raw and larval mortality-corrected (LMC) data, respectively (as indicated in the ‘Data’ column).

Significant parent–offspring correlations are outlined in boldface type.

Parent Offspring n Data r F1,68 P h²± SE

Paternal grandfathers F2fathers 70 Raw 0.386 11.902 0.001 0.92 ± 0.260

LMC 0.330 8.308 0.005 0.80 ± 0.272

Paternal grandmothers F2fathers 70 Raw –0.008 0.005 0.946 –0.01 ± 0.304

LMC 0.039 0.104 0.748 0.09 ± 0.282

Maternal grandfathers F2mothers 70 Raw 0.392 12.338 <0.001 0.78 ± 0.220

LMC 0.390 12.176 <0.001 0.70 ± 0.199

Maternal grandmothers F2mothers 70 Raw 0.138 1.324 0.254 0.32 ± 0.290

LMC 0.092 0.577 0.450 0.24 ± 0.332

F2fathers F3offspring 70 Raw –0.057 0.221 0.640 –0.08 ± 0.212

LMC 0.109 0.821 0.368 0.16 ± 0.173

F2mothers F3offspring 70 Raw –0.053 0.191 0.664 –0.07 ± 0.208

LMC 0.006 0.002 0.962 0.01 ± 0.172

Paternal grandparents F2fathers 70 Raw 0.266 5.173 0.026 0.45 ± 0.191

LMC 0.240 4.155 0.045 0.38 ± 0.183

Maternal grandparents F2mothers 70 Raw 0.384 11.780 0.001 0.58 ± 0.170

LMC 0.355 9.777 0.003 0.50 ± 0.160

F2parents F3offspring 70 Raw –0.069 0.330 0.568 –0.06 ± 0.133

LMC 0.072 0.355 0.554 0.06 ± 0.107

(9)

perfect survivorship may have left him with an unrep- resentative sample, especially considering that female bias and mortality appear to be correlated in this species (Igarashi, 1963a). To date, the male bias inTigriopusand other closely related copepods (Battaglia, 1958; Battaglia

& Malesani, 1959) has not been satisfactorily explained.

The absence of maternal transmission in this study was surprising because it contradicts our earlier work (Voordouw & Anholt, 2002b). The mother–offspring heritability of sex tendency in Voordouw & Anholt (2002b) was 1.24 ± 0.400 (recalculated using the cumulative standardized normal curve) and is comparable to the father–offspring heritabilities in Table 4. In Voordouw & Anholt (2002b), the sample size was small (17 mother–offspring pairs), and we could not rule out maternal inheritance of sex-biased mortality (i.e. survivorship was 65% in both generations). In the present study, the maternal grandmothers had low survivorship, sex-biased mortality and the highest (although nonsignificant) mother–offspring heritability estimate. This suggests maternal transmission of sex-specific viability differences and highlights the importance of minimizing larval mortality in this study. In addition, the lack of maternal transmission suggests that cytoplasmic sex-ratio distorters such asWolbachiado not play a role in the sex determination ofT. californicus.

In this study, the sex-ratio trait was paternally transmitted for both the paternal and the maternal F₁cross but not for the F2 cross (Fig. 3). Differences in the mating protocol between F₁and F₂crosses included precopulatory mate-guarding, age of fertilization and parity of the egg sac. In an earlier study, we found no effect of precopulatory mate-guarding on the primary sex ratio (unpublished data) but it is possible that the lack of mate-guarding in the F₂cross somehow inhibited paternal transmission of the sex-ratio factor (although we have no idea how this would occur). Volkmann-Rocco (1972) found that delayed fertilization caused females to increase the proportion of males in three species of Tisbe(a closely related genus of harpacticoid copepods). If this behaviour exists in T. californicusfemales, variance in the age of fertilization could compromise the father–offspring covariance in the primary sex ratio. If there are parity effects on the primary sex ratio (i.e. due to sperm or maternal aging), differences in egg sac parity between the three generations would likewise lower the parent–offspring covariance. For all the grandmothers in the F1cross, we used their first egg sac to produce the F₂generation. In the F₂cross by contrast, some of the mothers had already produced a sterile egg sac before they were mated and so we used their second egg sac to produce the F₃generation. Although parental age and clutch parity were more variable in the F3than the F2

generation these variables were, presumably, the most variable in the F1 generation where they comprised a haphazard sample from the field population. More likely, paternal transmission in the F2cross was compromised by poor survivorship in the F3generation (Table 1).

Paternal transmission of the sex-ratio is well known from two species of parasitoid wasp (Werren & Stoutha- mer, 2003); it was first discovered in N. vitripennis (Werrenet al., 1981) and was subsequently characterized in T. kaykai (Stouthamer et al., 2001). In these two species of parasitoid wasp, a supernumerary chromosome (B-chromosome) in the male’s sperm destroys the paternal genome (except themselves) shortly after fertilization. In the haplo–diploid Hymenoptera, males develop from unfertilized eggs and females develop from fertilized eggs. The destruction of the paternal genome by the B-chromosome results in a haploid embryo that then develops into a male (Werren & Stouthamer, 2003).

Haploid males produce sperm via mitosis whereas diploid females produce eggs via meiosis. B-chromosomes are lost during meiosis but not during mitosis (Werren &

Stouthamer, 2003). Hence from the perspective of the B-chromosome, destruction of the paternal genome is adaptive because it creates a mitotic lineage through which it can be transmitted to the next generation.

Paternal inheritance of a sex-ratio trait has also been reported in a polychaete worm, Ophryotrocha labronica (Premoliet al., 1996), and in the European fairy shrimp, B. schaefferi(Beladjal et al., 2002). Premoli et al. (1996) speculated that the sex-determining mechanism in O. labronica is a combination of female heterogamety and a polygenic system that modifies the sex of the embryo following fertilization. However, it is not clear to us why such a system would not generate a mother–

offspring correlation as the polygenic modifiers are presumably transmitted through the mother as well.

Beladjalet al.(2002) showed that the sperm ofB. schaefferi contains 10 autosomes and anywhere from one to three B-chromosomes. They suggested that the paternally transmitted sex-ratio trait in B. schaefferi is associated with the presence of B-chromosomes but did not explain how B-chromosomes would create such a pattern of inheritance in a diplo–diploid organism.

Associations between B chromosomes and sex ratios have been found in other organisms (Henderson, 1988;

Lopez-Leon et al., 1996; Vicente et al., 1996). In the ladybird beetle, Exochomus quadripustulatus, B-chromosome frequency was positively correlated with the adult proportion of females across 14 populations (Henderson, 1988). In the fish,Astyanax scabripinnis, B-chromosomes were more common in females than males and populations with a high prevalence of these B-chromosomes had female-biased sex ratios (Vicenteet al., 1996). In the grasshopper, Eyprepocnemis plorans, high B-chromosome load in field-mated parents was associated with a male- biased sex ratio in the offspring but this sex-ratio distortion was not detected in lab-mated individuals (Lopez-Leonet al., 1996). Again, for all of these examples, it is not clear how B-chromosomes and the observed sex-ratio bias are related.

One possibility is that these B-chromosomes carry a major sex-determining gene so that all individuals

(10)

with one or more B-chromosome develop into one sex (e.g. males) and all individuals who lack a B-chromosome develop into the other sex (e.g. females). We further envision a transmission rate that is some function of the B-chromosome load in the father.

Fathers with higher B-chromosome loads transmit more of them to their offspring and consequently have more sons. Barring some sort of fitness advantage, it is not clear how this male-determining B-chromosome could invade and replace an existing sex chromosome system (Camacho et al., 2000). However, we believe that the mechanism described above would account for the paternal inheritance pattern documented in this study and the male biased primary sex ratios commonly observed in T. californicus and other harpacticoid copepods.

Future research efforts will concentrate on karyotyping T. californicusto check for the presence of heteromorphic sex chromosomes (re: Ar-Rushdi, 1963) and to karyotype the sperm to test whether T. californicus males contain B-chromosomes (re: Beladjalet al., 2002) and how these B-chromosomes affect sex determination in their crusta- cean host.

Acknowledgments

This study was supported by Natural Sciences and Engineering Research Council (NSERC) of Canada research Grant OGP0138090 to B. R. Anholt and an NSERC PGS-B Scholarship to Maarten J. Voordouw (May 2002–April 2004). Eve thanks B. E. S. Robinson and N. Joo for their assistance and encouragement.

References

Ar-Rushdi, A.H. 1958. The polygenic basis of sex-ratio in Tigriopus. Proceedings of the Xth International Congress on Genetics, vol. 2, pp. 8–9, McGill University, Montreal, Canada.

Ar-Rushdi, A.H. 1963. The cytology of achiasmatic meiosis in the femaleTigriopus(Copepoda).Chromosoma13: 526–539.

Battaglia, B. 1958. Selezione e rapporto sessi nel Copepode HarpacticoideTisbe gracilis(T. Scott).Rend. Acc. Naz. Lincei Ser.

VIII24: 327–335.

Battaglia, B. & Malesani, L. 1959. Ricerche sulla determinazione del sesso nel CopepodeTisbe gracilis(T. Scott).Boll. Zool.26:

423–433.

Beladjal, L., Vandekerckhove, T.T.M., Heyrman, J., de Caesemae- ker, J. & Mertens, J. 2002. B-chromosomes and male-biased sex ratio with paternal inheritance in the fairy shrimpBranchipus schaefferi(Crustacea, Anostraca).Heredity88: 356–360.

Belser, W.L. 1959. The research frontier: where is science taking us?Sat. Rev.48: 58.

Bull, J.J. & Charnov, E.L. 1988. How fundamental are Fisherian sex ratios?Oxf. Surv. Evol. Biol.5: 96–135.

Bulmer, M.G. & Bull, J.J. 1982. Models of polygenic sex determination and sex ratio control.Evolution36: 13–26.

Burton, R.S. 1985. Mating system of the intertidal copepod Tigriopus californicus.Mar. Biol.86: 247–252.

Camacho, J.P., Sharbel, T.F. & Beukeboom, L.W. 2000.

B-chromosome evolution.Philos. Trans. R.. Soc. Biol. Sci.355:

163–178.

Carvalho, A.B., Sampaio, M.C., Varandas, F.R. & Klaczko, L.B.

1998. An experimental demonstration of Fisher’s principle:

evolution of sexual proportion by natural selection. Genetics 148: 719–731.

Cosmides, L.M. & Tooby, J. 1981. Cytoplasmic inheritance and intragenomic conflict.J. Theor. Biol.89: 83–129.

Dawkins, R. 1976. The Selfish Gene. Oxford University Press, Oxford.

Egloff, D.A. 1966. Biological Aspects of Sex Ratio in Experimental Field Populations of the Marine Copepod Tigriopus Californicus.

Stanford University, Stanford, California.

Falconer, D.S. 1989. Introduction to Quantitative Genetics. Wiley and Sons, New York.

Fisher, R.A. 1930. The Genetical Theory of Natural Selection.

Clarendon Press, Oxford.

Hamilton, W.D. 1967. Extraordinary sex ratios.Science156: 477–

488.

Henderson, S.A. 1988. A correlation between B chromosome frequency and sex ratio inExochomus quadripustulatus. Chro- mosoma96: 376–381.

Howard, H.W. 1942. The genetics ofArmadillidium vulgareLatr.

II. Studies on the inheritance of monogeny and amphogeny.

J. Genet.44: 143–159.

Hurst, L.D. 1993. The incidences, mechanisms and evolution of cytoplasmic sex ratio distorters in animals.Biol. Rev.68: 121–

193.

Igarashi, S. 1963a. On the stability of sex ratio and the prediction of its abnormality depending on the color of the ovisac in Tigriopus japonicusMori.Sci. Rep. Tohoku Univ. Fourth Ser. (Biol.) 29: 83–90.

Igarashi, S. 1963b. The primary sex ratio of a marine copepod, Tigriopus japonicus.Sci. Rep. Tohoku Univ. Fourth Ser. (Biol.)29:

73–81.

Jaenike, J. 2001. Sex chromosome meiotic drive.Ann. Rev. Ecol.

Syst.32: 25–49.

Jiggins, F.M., Hurst, G.D.D., Dolman, C.E. & Majerus, M.E.N.

2000. High-prevalence male-killingWolbachiain the butterfly Acraea encedana.J. Evol. Biol.13: 495–501.

Juchault, P., Rigaud, T. & Mocquard, J.-P. 1993. Evolution of sex determination and sex ratio variability in wild populations of Armadillidium vulgare(Latr.) (Crustacea, Isopoda): a case study in conflict resolution.Acta Oecol.14: 547–562.

Legrand, J.J., Legrand-Hamelin, E. & Juchault, P. 1987. Sex determination in Crustacea.Biol. Rev. Cam. Phil. Soc.62: 439–

470.

Lewis, D. 1941. Male sterility in natural populations of herma- phroditic plants: the equilibrium between females and her- maphrodites to be expected with different types of inheritance.New Phyt.40: 56–63.

Lopez-Leon, M.D., Cabrero, J. & Camacho, J.P.M. 1996.

Negatively assorted gamete fertilization for supernumerary heterochromatin in two grasshopper species.Heredity76: 651–

657.

O’Neil, S.L., Hoffmann, A.A. & Werren, J.H. 1997. Influential Passengers: Inherited Microorganisms and Arthropod Reproduction.

Oxford, University Press, Oxford.

Premoli, M.C., Sella, G. & Berra, G.P. 1996. Heritable variation of sex ratio in a polychaete worm. J. Evol. Biol.

9: 845–854.

(11)

Rigaud, T. 1997. Inherited microorganisms and sex determination of arthropod hosts. In: Influential Passengers: Inherited Microorganisms and Arthropod Reproduction(S. L. O’Neil, A. A.

Hoffmann & J. H. Werren, eds), pp. 81–101. Oxford University Press, Oxford, U.K.

Roff, D.A. 1986. The genetic basis of wing dimorphism in the sand cricket, Gryllus firmus and its relevance to the evolution of wing dimorphism in insects. Heredity 57:

221–231.

Roff, D.A. 1997.Evolutionary Quantitative Genetics. Chapman and Hall, New York.

Rousset, F., Bouchon, D., Pintureau, B., Juchault, P. & Solignac, M. 1992. Wolbachia endosymbionts responsible for various alterations of sexuality in arthropods.Proc. R. Soc. Lond. Ser. B Biol. Sci.250: 91–98.

Stouthamer, R., Breeuwer, J.A.J. & Hurst, G.D.D. 1999.

Wolbachia pipientis: microbial manipulator of arthropod reproduction.Ann. Rev. Micro.53: 71–102.

Stouthamer, R., van Tilborg, M., de Jong, J.H., Nunney, L. &

Luck, R.F. 2001. Selfish element maintains sex in natural populations of a parasitoid wasp.Proc. R. Soc. Lond. Ser. B Biol.

Sci.268: 617–622.

Taylor, D.R. 1990. Evolutionary consequences of cytoplasmic sex ratio distorters.Evol. Ecol.4: 235–248.

Varandas, F.R., Sampaio, M.C. & Carvalho, A.B. 1997. Herit- ability of sexual proportion in experimentalsex-ratiopopula- tions ofDrosophila mediopunctata.Heredity79: 104–112.

Vicente, V.E., Moreira-Filho, O. & Camacho, J.P.M. 1996. Sex- ratio distortion associated with the presence of a B chromosome inAstyanax scabripinnis(Teleostei, Characidae).Cyt. Cell Gen.74: 70–75.

Volkmann-Rocco, B. 1972. The effect of delayed fertilization in some species of the genusTisbe (Copepoda, Harpacticoida).

Biol. Bul.142: 520–529.

Voordouw, M.J. & Anholt, B.R. 2002a. Environmental sex determination in a splash pool copepod.Biol. J. Linn. Soc.76:

511–520.

Voordouw, M.J. & Anholt, B.R. 2002b. Heritability of sex tendency in a harpacticoid copepod, Tigriopus californicus.

Evolution56: 1754–1763.

Weeks, A.R., Reynolds, K.T. & Hoffmann, A.A. 2002.Wolbachia dynamics and host effects: what has (and has not) been demonstrated?TREE17: 257–262.

Werren, J.H. 1997. Biology ofWolbachia.Ann. Rev. Entom. 42:

587–609.

Werren, J.H., Skinner, S.W. & Charnov, E.L. 1981. Paternal inheritance of a daughterless sex ratio factor.Nature293: 467–

468.

Werren, J.H. & Stouthamer, R. 2003. PSR (paternal sex ratio) chromosomes: the ultimate selfish genetic elements.Genetica 117: 85–101.

Received 30 October 2004; revised 8 February 2005; accepted 15 February 2005